I'd like to get some advice and comment from you CFD On-line peoples on the followings:

1. How large the CFD market size is in worldwide? 2. How is the growth rate in anual? 3. What would be the furture CFD market? 4. In the view of technology trend, which direction would be most prospected?

Thank you for your attention, and hope for reply.

In general, CFD ( Computational Fluid Dynamics) deals with Fluid Dynamics related problems through the use of computer simulation methods ( solving a set of equations, such as Navier-Stokes Equations with proper turbulence models ). The field is un-limited. It covers areas such as, aircraft, jet engine, missiles and rocket, combustors, automobile, internal combustion engine, combustors, fan, pumps, air conditioning, ship, submarine, electronic cooling, radiator, heat exchanger, metal casting, bio-fluid mechanics, nuclear explosion, lunar landing, space shuttle launching, etc.....etc.... So, the CFD field is very, very, very big !!! In 60's, people used IBM main frame computer to perform CFD activities. In 70's, mini-computers became available. In 80's, super-computers and workstations were very attractive. In 90's, workstations and PC's become very popular. As the price of computers drops and the performance of the PC increases, In the next century, PC will be widely used in simulation. The total number of workstations and PC's must be a very big number. CFD is not a product, or a software, or service. CFD is a way of doing thing. The field is so big that you don't have to worry about it , today or next century. The hard part is how to do it better. And you will find those information somewhere in this site, or forum ,from time to time.

I'm not sure what you mean with "CFD market". If you mean the market for commercial CFD codes:

Fluent's sales the first quarter of 1998 was around 10 million US dollars, with a gross profit margin of 25%. Sales was up about 20% compared to the first quarter last year. I don't know exactly how big Fluent's market share is, but it is a significant share - anyone care to guess how many percent they have?

Allow me to contribute the following: As of end FY 97, total code sales by general purpose CFD code suppliers was around $85million. If one adds the 'niche' code sales as well as those of 'pre' and 'post' processing software, this number goes up to around $100million. Fluent's market share in the general purpose CFD market is 44% worldwide, 47% in the USA, 57% in Europe and 31% in Asia Pasific. The market share of the other two main CFD vendors are around 19% and 14% worldwide. I hope that this information is useful.

Interesting numbers. Makes me wonder how big the different markets (USA/Europe/Asia) are. Have you got any data on that?

Assuming that Fluent company has more than just one product(5 assumed) and assuming that each license ( or copy ) costs $20,000 per year and on the average each user company has 5 licenses ( I know one company has around 15 licenses), then on the average, the total number of user company is around 20. I am sure that the number you quoted also includes the consulting activities ( in addition to the software sale ). Based on the National CFD conferences held in Taiwan alone, there are at least 50 to 100 dedicated CFD researchers ( at university professor and PhD student level) working in various CFD field. Most of them are writing their codes. ( There are at least a couple of reasons for this, one is the high cost of a commercial code, the other is the need to modify the code for research and teaching purposes. The student must be able to access the source code in order to try out his new ideas. They are also paper publication oriented. The schemes implemented in a commercial code normally are out-dated by the time when it becomes available as an option. ) It looks like that research and teaching institution are not ideal place for commercial codes.

The sales figure I mentioned was for for what they define as "software sales" the first quarter of 1998. You have to multiply it with 4 to get the annual figure. I don't know if they include consulting in this "sofware sales".

With new unstructured methods, advanced non-linear turbulence models (reynolds stress tensor expressions that cover several pages when you write them out), complex geometries and professional demands on post-processing etc.... I am afraid that a large part of academia will be forced to turn to commercial codes eventually. In order to maintain and keep a modern CFD code current you will have to have team of professional software engineers working on it full time - snippets of code can't just be handed down from graduate student to graduate student like it is used to be, and still is at most places. Not many University Professors will have the interest, time and knowledge necessary to maintain a modern CFD code. At least that is what I belive. I'm not saying this is a good thing! Of course there will be exceptions and more "pure" scientist will naturally have their own codes.

Note also that this development is in the interest of the commercial code vendors and I'm sure that they will offer very good educational licenses, access to parts or of source code, higher modularity etc. Perhaps we will even see a CFD "linux" - that would be nice!

I am sure that the CFD field is big enough and wide open to the commercial CFD code developer. Last year, when I mentioned the unstructured mesh and hybrid mesh used in one of the commercial code to a friend of mine at NASA, he said that they had done studies back in 80's and were aware of the problems. If you use only the triangular mesh near the wall, you are going to have very hard time to get consistent answers unless the mesh is uniform. The hybrid mesh is the logical way to handle the problem, that is to use densely packed prism mesh near the wall. But, to make it practical and easy-to-use for any geometry is a challenging job. Access to the advanced CFD source codes is not a problem when you are working with government and contractors of the projects, as long as you have signed the paper and followed the rules. For the need to know purpose, CFD researchers normally have no problems. You must have the access to the source code. Decoding of a source code normally takes anywhere from 6 months to one year, because these codes normally don't come with good documents. I still say that it is faster to write your own code than to decode someone else's code and then to make modifications. At this stage of commercial code development, it is not a good idea to give away the code ( free of charge to university or something like that ). Fluid dynamic problems are always very difficult to solve. Computer hardware companies had tried before to give away computers to schools, hoping that they will continue to use the same brand after graduation. Unfortunately, with the fast moving technology, students are attracted to the newer, faster and cheaper computers on the market. Same thing will happen to the commercial codes. Users will have the freedom to select any codes with better features in it, faster speed, better accuracy, cheaper cost, more user-friendly.

Ferit, what are the other 2 companies you refer to above? Are they general purpose or 'niche'?

yes, i do have numbers on the size of various geographic markets. they are roughly equal, with the US being slightly larger. the reason for the japanese market being so large is due to japanese prices.

robin, they are general purpose codes. i have left out the niche players. however, the largest 'niche' player turns over about $10mil. i hope that this answers your question.

Howdy,

I agree with the $100 million gross per year worldwide number.

However, the breakdown you give does not agree with what I've seen. To be a little more specific, the worldwide breakdown that I recently saw was this:

Fluent 30% CFX-AEA 25% Star-CD 20% Other 25%

Not much different I guess, but close. Apparently, the gross amount has been growing about 25-30% per year for the last 10 years as well. So, it is definitely a growth market.

Dan W.

Additional info,

Ferit asked me where I saw the breakdown which was in my previous posting. It was in a report, which is sitting in our office here, that was done by the sales team. The references in there were as follows:

Now that both Fluent and ASC have been bought out by public companies, their financial information is in the public domain. We have reports laying around in the office here. No big mystery there.

As far as the Star-CD number goes, that was just based on information provided by someone from Computational Dynamics, to a manager here.

The growth numbers are based on internal studies done at AEA.

Just curious ferit, where did you get your information about the "other companies", which you did not mention the names of. Only that they are commercial codes.

Dan.

Did Eric send you that paper I found on non-reflective B.C.'s?